Iournal of Agricultural OCTOBER 1994 VOLUME 42, NUMBER 10
andFd Chemistrv
0Copyright 1994 by the American Chemical Society
SHORT COMMUNICATIONS Analysis of Wines for Resveratrol Using Direct Injection High-pressure Liquid Chromatography with Electrochemical Detection Keywords: Resveratrol; wine; HPLC;electrochemical detectors; phenolic stilbenes; quercetin INTRODUCTION
Resveratrol [(E)-3,4’,5-trihydroxystilbenel is produced by grapevines (Langcake, 1981) and occurs in grapes (Jeandet et al., 1992)and wines (Siemann and Creasy, 1992;Lamuela-Raventos and Waterhouse, 1993;Pezet et al., 1994) produced from these grapes. It has been speculated that this substance may have positive effects on human cardiovascular health, acting to lower serum lipid levels (Siemann and Creasy, 1992). Recently, details on the development and application of a method for analysis of resveratrol in wines using HPLC with monitoring at several wavelengths of W light were published (Lamuela-Raventos and Waterhouse, 1993). The method involved reduction of volume of 500 mL of wine to approximately 450 mL to remove ethanol, extraction with chloroform to remove interfering substances, extraction with ethyl acetate to isolate the resveratrol, evaporation of the ethyl acetate, and reconstitution of the residue in 10 mL of aqueous acetonitrile followed by filtration of the sample through 0.45pm Teflon filters prior to HPLC analysis using an octadecyl reversed-phase column. The authors noted that recovery of resveratrol was somewhat modest (45%) but that the sensitivity of the method was limited to about 0.05 mg/L by the presence of interferences rather than recovery. The method proved to be successful, especially in decreasing the time needed for analysis compared to one previous method (Siemann and Creasy, 1992)and in sensitivity compared to another (Jeandet et al., 1992;Langcake and Pryce, 1976). Very recently, the application of HPLC with fluorescence detection to the analysis of wines and grape berries for resveratrol and pterostilbene has been described (Pezet et al., 1994). Fluorescence detection allowed rapid analysis of wines for the two target 0021-8561/94/1442-2077$04.50/0
compounds. The method possesses sufficient selectivity and sensitivity to allow direct analysis of wines without prior sample preparation. We have been interested in resveratrol and other phenolic stilbenes elucidated in various plant species because of their possible role in fungus resistance of certain species of woody plants (Schultz et al., 1992). We have found that reversed-phase HPLC with electrochemical detection (EC)is very useful for determining phenolic substances such as resveratrol and other hydroxylated stilbenes in biological systems including plants and products deriving from plants. Electrochemical HPLC detectors are relatively well-known in the field of neurochemistry but are used less often in other areas. EC detectors are selective detectors. When operated in the oxidative mode, they respond well to phenols and are capable of parts-per-billion or even lower detection levels for these substances. This paper demonstrates that reversed-phase HPLC with electrochemical detection allows analysis of wines for resveratrol by direct injection with sensitivities at least as good as those based on fluorescence detection. This level of selectivity and sensitivity was obtained using a simple isocratic HPLC system. MATERIALS AND METHODS Resveratrol (Figure 1)was synthesized in this laboratory using a method developed for synthesis of the tetrahydroxylated stilbene piceatannol (Bajaj et a l . , 1987). Quercetin was purchased from Sigma Chemical Co. (St. Louis, MO), and 2,6di-tert-butyl-4-methylphenol (BHT)was obtained from Aldrich Chemical Co. (Milwaukee, WI). Two liquid chromatographs were used in this study. One chromatograph has gradient pumps, a variable-wavelength W detector with provision for stopped-flow scanning (200-800 0 1994 American Chemical Society
Short Communications
2078 J. Agric. Food Chem., Vol. 42,No. 10, 1994 OH
Ho\
0
100
Resveratrol .
OOH f 0 l
o
H
Quercetin
'""'b ?H
t-Bu
BHT
CH3
Figure 1. Structures of BHT, quercetin, and resveratrol. nm) of the W spectra of components being analyzed, and a variable fluorescence detector (Perkin-Elmer, Norwalk, CT). This chromatograph was used t o confirm quantitation of resveratrol in some of the red wine samples and to obtain U V spectra of synthetic resveratrol and the component of the Pinot noir wine which eluted with the retention time of resveratrol. All quantitative determinations were obtained using a very simple modular isocratic liquid chromatographassembled from various components. A Simplex micrometering pump (LDC/ Milton Roy, Riviera Beach, FL), a coiled stainless-steel tubular pulse dampener, a pressure gauge (Alltech Associates, Deerfield, IL), a six-port injection valve (Valco Instruments, Houston, TX), an electrochemical detector (Model LC-3, Bioanalytical Systems, Lafayette, IN), and a strip-chart recorder (Linear Instruments, Irvine, CA) comprised this chromatograph. Samples were quantitated by measuring peak heights manually. The electrochemical detector included a glassy carbon working electrode and a silverkilver chloride reference electrode (BioanalyticalSystems). Columns used were all 150 mm long, 4.6 mm i.d., and were packed in our laboratory. Two columns of different chemistry were used. These were an octadecyl column (NucleosilCIS,5 pm, Macherey-Nagel,Duren, Germany) and a benzyldimethylsilyl bonded phase. The benzyldimethyl packing was prepared in this laboratory following procedures previously described using irregular silica gel with a particle size of 7 f 2 pm (McMurtrey, 1988). The Nucleosil Cl8 column was used for all quantitative determinations. Mobile phases were 0.05 m o m NHdHzP04 (E. Merck, Darmstadt, Germany) in 2570 (vh) aqueous acetonitrile (Optima grade, Fisher Chemical, Fair Lawn, NJ). Mobile phases were prepared using deionized water. Extraction cartridges used were from Worldwide Monitoring (Horsham, PA). These consisted of CISgroups bonded to silica gel and were endcapped: size, 500 mg; adsorbent, sample volume, 6 mL. Wines were purchased from commercial suppliers in the area. Wines selected include the following: (1)Classic Red, Deer Valley Vineyards, Gonzalez, CA (nonvintage blend of Barbera, Pinot noir, Zinfandel, and Ruby Cabernet); (2) 1991 California Zinfandel, Sutter Home Winery, St. Helena, Napa County, CA, (3) 1991 Concha y Toro, 75% Cabernet Sauvignod 25% Merlot, Rapel, Vina Concha y Toro, SA, Chile; (4) 1990 Eye of the Swan Cellars California Pinot noir, Sebastiani Vineyards, Sonoma, CA; (5) 1986 Northern Sonoma Cabernet Sauvignon of California, The Reserve Cellars of Ernest and Julio Gallo, Northern Sonoma County, CA (6) Blossom Hill Reserve Collection Chardonnay, Madera, CA, ( 7 ) 1992 Beaujolais-Villages Jadot, Beaujolais-Villages Controlee, Louis Jadot, Beaune, Cote-D'Or, France; (8) 1992 Saint-Louis Beaujolais, appellation Beaujolais Controlee, Barton & Guestier, Blanquefort, France (burgundy red wine); (9)Classic White, premium California table wine, Deer Valley Vineyards, (nonvintage blend of 60% French Colombard, 15% Chenin blanc, 7% Muscat, and 18% white Riesling); (10) 1991 Sauvignon Blanc of California, Ernest and Julio Gallo, Modesto, CA and
0
I
013
014
0.5
,
0.6
0:7
0:s
03
i
volts vs Ag/AgCl Figure 2. Relative response (peak height) of resveratrol (R), and butylated hydroxytoluene(BHT)at various quercetin (Q), voltages of the working electrode relative t o a silver/silver chloride reference electrode. Responses were normalized to the signal at 1.00 V applied potential. (11)Gallo Cocktail Pale Dry Sherry of California, Ernest and Julio Gallo, Modesto, CA.
RESULTS AND DISCUSSION Electrochemical detectors are operated with the working electrode at a potential at which analyte molecules undergo electrochemical reactions. The specific potential t h a t is used depends on the structures of t h e molecules being analyzed. Resveratrol; quercetin, a natural product; a n d BHT, a synthetic phenol antioxid a n t often added to various foods, (see Figure 1 for structures) were analyzed at applied voltages from $0.300 to +1.000 V vs Ag/AgCl. The results a r e presented in Figure 2. Both resveratrol a n d quercetin give rather broad responses, apparently because of the presence in these molecules of more t h a n one phenolic hydroxyl. In contrast, t h e applied voltage-response curve from BHT is much more narrow. The response of resveratrol at relatively low applied voltages is a n advantage since a low voltage will give a relatively low background signal a n d reduces the response of interfering substances. In effect, t h e use of a low working voltage increases the selectivity of the detector. For example, analysis of a mixture of resveratrol a n d BHT at 0.5 V would give a strong signal for resveratrol but no signal for BHT. If 0.9 V or higher were used, both compounds could be monitored. We chose 0.500 V vs Agl AgCl for analysis of resveratrol. Detector response to standard samples of resveratrol (peak height) was linear over the range tested (0.108.00 m g L ) with a correlation coefficient of 0.999. The region of linearity extends to concentrations less t h a n 0.10 mg/L. Quantitation could be extended to at least 0.01 mg/L and perhaps lower by simply using a more sensitive attenuation of t h e EC detector. The limit of detection of resveratrol in wine appears to be approximately 1 ppb (ug/L) for t h e method as described. Even greater sensitivities might be obtained by concentrating the analyte molecules. In this context, we evaluated a Cis-based extraction cartridge a n d found t h a t it allowed extraction of 1.00 mg/L resveratrol from wine with a recovery of 89 & 1%.Extraction cartridges can be employed to concentrate resveratrol i n wine samples (Mattivi, 1993) without recourse to rotary evaporation, which h a s been reported to lead to decomposition of hydroxylated stilbenes (Pezet et al., 1994). Cartridge isolation also could be used to isolate other components of wine that have lipophilicities similar to that of resveratrol. The extraction efficiencies of various
J. Agric. Food Chem., Vol. 42, No. 10, 1994 2079
Short Communications
nm Figure 3. W spectra of (A) resveratrol in Pinot noir wine and (B) synthetic resveratrol. Spectra were obtained by stopped-flow scanning of the HPLC eluent.
wine components eluting after about 10 min were approximately equivalent to that found for resveratrol. We were somewhat surprised by the apparent instability of resveratrol during concentration by rotary evaporation at 40 "C as described by Pezet et al. (1994). We have isolated resveratrol from the heartwood of decay-resistant trees. We have found that the resveratrol present in heartwood is stable in dried, ground heartwood samples for a t least 2 years. In addition, we have found that hydroxystilbenes are usually thermally stable. Part of the instability of resveratrol during isolation by rotary evaporation may arise from catalysis by the glass surface or by light-mediated reactions. Certainly, analyte loss during concentration of partsper-million or parts-per-billion samples is always a possible source of analytical inaccuracy and must be guarded against. There appeared to be little if any interfering materials in the chromatograms with electrochemical detection. This conclusion was reached as a result of several factors. First, the electrochemical detector is a selective detector. Second, the resveratrol peaks were very symmetrical. They show no evidence of overlapping or coeluting substances. Third, the stopped-flow W spectra for synthetic resveratrol and for the resveratrol peak obtained from analysis of 100 p L of the Pinot noir wine were essentially identical (Figure 3). Fourth, we isolated resveratrol from the Pinot noir wine with the same CIS column used during quantitative determinations. This material was then analyzed using the EC detector connected to a benzyldimethylsilyl column. Only one peak eluted with a retention time which matched exactly that of resveratrol. Fifth, the three wines with the highest resveratrol levels (the two Beaujolais wines and the Pinot noir) were analyzed using a fluorescence detector. Resveratrol levels by these two techniques were equivalent within experimental error. Eleven wines were analyzed for resveratrol content. The results are given in Table 1, with representative chromatograms displayed in Figure 4. As expected, white wines contained very little resveratrol. In contrast, the Pinot noir tested contained resveratrol at a level of 5.01 f 0.25 mg/L. In general, red wines from vines originating in the Burgundy region of France (Gamay, Pinot noir) appear to contain relatively high levels of resveratrol. While absolute amounts are different from the values reported in this paper, others have reported similar trends. Lamuela-Raventos and Waterhouse (1993) found higher levels in Pinot noir wine (from 0.21 to 0.68 mg/L) than in Zinfandel wines (0.06 and 0.11 mg/L) and Cabernet Sauvignon (~0.050.09 mg/L). All wines in this study were from the same area of California and were the same vintage (1989). Pezet et al. (1994) analyzed wines produced in Switzer-
Table 1. Concentration of Resveratrol in Selected Wines wine oriein concna Ime/L) California 50.01 (11)dry sherry 50.02 (6) Chardonnay California California 50.02 (9) generic white 50.02 California (10)Sauvignon blanc 0.99 f 0.08 California (5)Cabernet Sauvignon California 1.38 & 0.18 (2) Zinfandel 1.56 & 0.08 (3) Cabernet Sauvignofierlot Chile California 2.74 z t 0.08 (1)generic red 3.27 f 0.14 France (8) Beaujolais (Gamay) 3.55 f 0.06 France (7)Beaujolais (Gamay) 5.01 f 0.25 California (4) Pinot noir
aMean & sample standard deviation, from three or more determinations.
ll 'I 1I
0
A 100 nAFS
11
Resveratrol
10
20
30
MINUTES Figure 4. Representative chromatograms of analysis of wines: (A) Chardonnay, wine 6, resveratrol content 50.02 mgl L; (B)Cabernet Sauvignon, wine 5, resveratrol content 0.99 f 0.08 m a ; (C) Pinot noir, wine 4,resveratrol content 5.01 f 0.25 m g L Sensitivity in (A) is 5 times greater than in (B) and 10 times greater than in (C).
land. These authors found the highest resveratrol levels in a Gamay wine (1.48 mg/L) with levels in a Pinot and a Gamaret wine of 0.67 and 0.43 mg/L, respectively. Data available a t this time are insufficient to allow complete assessment of the levels of resveratrol in commercial wines to be made. Knowledge of year-toyear variation of resveratrol levels in wines produced in one area, as well as variation of resveratrol content as a result of geographical area, would allow better estimates to be made of the amounts of resveratrol that average wine drinkers ingest. When such data become available, more meaningful assessment of the possible role that resveratrol in the diet may have on human health may be possible. The electrochemical detector can also be used to monitor wines for the presence of other phenolic substances. For example, the substance eluting with a retention time of 25 min in the chromatograms displayed in Figure 4 is primarily quercetin. We did not quantitate this material because there are substances which overlap. Relatively minor modification of the analytical methodology should allow quercetin (and other hydroxyflavanoids) to be quantitated.
2080 J. Agric. Food Chem., Vol. 42, No. 10, 1994
The analytical techniques described here allow analysis of wines for resveratrol without sample pretreatment. Method sensitivity is more than adequate to monitor levels of resveratrol a t considerably less than 1mg/L (1part per million). The method was developed using the most simple liquid chromatograph and therefore is relatively inexpensive. The method appears t o compete successfully with methods based on other means of detection in terms of convenience, sensitivity, and cost. ACKNOWLEDGMENT
We gratefully acknowledge the support of the Department of Chemistry and Biochemistry of the University of Southern Mississippi, the Forest Products Laboratory of Mississippi State University, and the U S . Department of Agriculture (Grant WUR3556726) for support of this research.
Short Communications
Langcake, P.; Pryce, R. J. The production of resveratrol by Vitis vinifera and other members of the Vitaceae as a response to infection or injury. Physiol. Plant Pathol. 1976, 9, 7786. Mattivi, F. Solid-phase extraction of trans-resveratrol from wines for HPLC analysis. 2.Lebensm. Unters. Forsch. 1993, 196, 522-525. McMurtrey, K. D. Reaction of silica gel with trimethylsilyl donors under conditions useful for end-capping HPLC bonded phase packings. J . Liq. Chromatogr. 1988,11,33753384. Pezet, R.; Pont, V.; Cuenat, P. Method to determine resveratrol and pterostilbene in grape berries and wines using highperformance liquid chromatography with highly sensitive fluorometric detection. J . Chromatogr. A 1994, 663, 191197. Schultz, T. P.; Boldin, W. D.; Fisher, T. H.; Nicholas, D. D.; McMurtrey, K. D.; Pobanz, K. Structure-fungicidalproperties of some 3- and 4-hydroxylated stilbenes and bibenzyl analogues. Phytochemistry 1992,31, 3801-3806. Siemann, E. H.; Creasy, L. L. Concentration of the phytoalexin resveratrol in wine. Am. J . Enol. Vitic. 1992, 43, 49-52.
LITERATURE CITED Received for review May 18, 1994. Accepted August 2, 1994. Bajaj, R.; Gill, M. T.; McLaughlin, J. L. Improved preparative synthesis of piceatannol (3,4,3’,5’-tetrahydroxy-trans-stilbene). Rev. Latinoam. Quim. 1987,18, 79-80. Jeandet, P.; Sbaghi, M.; Bessis, R. The production of resveratrol(3,4‘,5-trihydroxystilbene)by grape berries in different development stages. Am. J. Enol. Vitic. 1992,42, 41-46. Lamuela-Raventos, R. M.; Waterhouse, A. L. Occurrence of resveratrol in selected California wines by a new HPLC method. J . Agric. Food Chem. 1993,41, 521-523. Langcake, P. Disease resistance of uitis spp. and the production of the stress metabolites resveratrol, y-viniferin, a-viniferin and pterostilbene. Physiol. Plant Pathol. 1981,18,213-226.
Kenneth D. McMurtrey,’JJames MinnJ Kirk Pobanz3 and Tor P. Schultzs Department of Chemistry and Biochemistry, University of Southern Mississippi, Box 5043, Hattiesburg, Mississippi 39406, and Mississippi Forest Products Laboratory, P.O. Drawer FP, Mississippi State, Mississippi 39762-5724
’University of Southern Mississippi.
* Mississippi Forest Products Laboratory.
